def test_rebuild_old_chunk(self):
for c in self.chunks:
convert_to_old_chunk(
self._chunk_path(c), self.account, self.container, self.path,
self.version, self.content_id)
chunk = random.choice(self.chunks)
chunk_volume = chunk['url'].split('/')[2]
chunk_id = chunk['url'].split('/')[3]
chunk_headers, chunk_stream = self.blob_client.chunk_get(
chunk['url'], check_headers=False)
os.remove(self._chunk_path(chunk))
chunks_kept = list(self.chunks)
chunks_kept.remove(chunk)
conf = self.conf.copy()
conf['allow_same_rawx'] = True
rebuilder = BlobRebuilder(conf, service_id=chunk_volume)
rebuilder_worker = rebuilder.create_worker(None, None)
rebuilder_worker._process_item(
(self.ns, self.cid, self.content_id, chunk_id))
_, new_chunks = self.api.object_locate(
self.account, self.container, self.path)
new_chunk = list(new_chunks)
self.assertEqual(len(new_chunks), len(chunks_kept) + 1)
url_kept = [c['url'] for c in chunks_kept]
new_chunk = None
for c in new_chunks:
if c['url'] not in url_kept:
self.assertIsNone(new_chunk)
new_chunk = c
self.assertNotEqual(chunk['real_url'], new_chunk['real_url'])
self.assertNotEqual(chunk['url'], new_chunk['url'])
self.assertEqual(chunk['pos'], new_chunk['pos'])
self.assertEqual(chunk['size'], new_chunk['size'])
self.assertEqual(chunk['hash'], new_chunk['hash'])
new_chunk_headers, new_chunk_stream = self.blob_client.chunk_get(
new_chunk['url'])
chunk_data = b''.join(chunk_stream)
new_chunk_data = b''.join(new_chunk_stream)
self.assertEqual(chunk_data, new_chunk_data)
fullpath = encode_fullpath(self.account, self.container, self.path,
self.version, self.content_id)
self.assertEqual(fullpath, new_chunk_headers['full_path'])
del new_chunk_headers['full_path']
self.assertNotEqual(chunk_headers['chunk_id'],
new_chunk_headers['chunk_id'])
new_chunk_id = new_chunk['url'].split('/')[3]
self.assertEqual(new_chunk_id, new_chunk_headers['chunk_id'])
del chunk_headers['chunk_id']
del new_chunk_headers['chunk_id']
self.assertEqual(OIO_VERSION, new_chunk_headers['oio_version'])
del chunk_headers['oio_version']
del new_chunk_headers['oio_version']
self.assertEqual(chunk_headers, new_chunk_headers)
def _take_action(self, parsed_args):
if not self.distributed: # local
self.tool_conf['dry_run'] = parsed_args.dry_run
self.tool_conf['try_chunk_delete'] = \
parsed_args.delete_faulty_chunks
self.rebuilder = BlobRebuilder(self.tool_conf,
input_file=parsed_args.input_file,
logger=self.logger)
if self.distributed:
self.rebuilder.prepare_distributed_dispatcher()
else:
self.rebuilder.prepare_local_dispatcher()
for item, _, error in self.rebuilder.run():
if error is None:
status = 'OK'
else:
status = 'error'
yield (self.rebuilder.string_from_item(item), status, error)
def send_chunk_job(self, target, irreparable=False):
"""
Send a "content broken" event, to trigger the
reconstruction of the chunk.
"""
item = (self.api.namespace, target.cid, target.content_id,
target.chunk)
ev_dict = BlobRebuilder.task_event_from_item(item)
if irreparable:
ev_dict['data']['irreparable'] = irreparable
job = json.dumps(ev_dict)
self.error_sender.send_job(job)
self.error_sender.job_done() # Don't expect any response
class RawxRebuildCommand(SingleServiceCommandMixin, ServiceRebuildCommand):
tool_class = BlobRebuilder
columns = ('Chunk', 'Status', 'Errors')
def get_parser(self, prog_name):
parser = super(RawxRebuildCommand, self).get_parser(prog_name)
SingleServiceCommandMixin.patch_parser(self, parser)
# common
parser.add_argument(
'--rdir-fetch-limit',
type=int,
help='Maximum number of entries returned in each rdir response. '
'(default=%d)' % self.tool_class.DEFAULT_RDIR_FETCH_LIMIT)
if not self.distributed: # local
parser.add_argument('--dry-run',
action='store_true',
help='Display actions but do nothing. '
'(default=%s)' %
self.tool_class.DEFAULT_DRY_RUN)
parser.add_argument(
'--delete-faulty-chunks',
action='store_true',
help='Try to delete faulty chunks after they have been '
'rebuilt elsewhere. This option is useful if the chunks '
'you are rebuilding are not actually missing but are '
'corrupted. '
'(default=%s)' % self.tool_class.DEFAULT_TRY_CHUNK_DELETE)
return parser
def _take_action(self, parsed_args):
# common
self.tool_conf['rdir_fetch_limit'] = parsed_args.rdir_fetch_limit
if not self.distributed: # local
self.tool_conf['dry_run'] = parsed_args.dry_run
self.tool_conf['try_chunk_delete'] = \
parsed_args.delete_faulty_chunks
self.rebuilder = BlobRebuilder(self.tool_conf,
service_id=parsed_args.service,
logger=self.logger)
if self.distributed:
self.rebuilder.prepare_distributed_dispatcher()
else:
self.rebuilder.prepare_local_dispatcher()
for item, _, error in self.rebuilder.run():
if error is None:
status = 'OK'
else:
status = 'error'
yield (self.rebuilder.string_from_item(item), status, error)
def take_action(self, parsed_args):
SingleServiceCommandMixin.check_and_load_parsed_args(
self, self.app, parsed_args)
return super(RawxRebuildCommand, self).take_action(parsed_args)
class ChunkRebuildCommand(ItemRebuildCommand):
tool_class = BlobRebuilder
columns = ('Chunk', 'Status', 'Errors')
def get_parser(self, prog_name):
parser = super(ChunkRebuildCommand, self).get_parser(prog_name)
parser.add_argument(
'--input-file',
help='Read chunks from this file instead of rdir. '
'Each line should be formatted like '
'"container_id|content_id|short_chunk_id_or_position".')
if not self.distributed: # local
parser.add_argument('--dry-run',
action='store_true',
help='Display actions but do nothing. '
'(default=%s)' %
self.tool_class.DEFAULT_DRY_RUN)
parser.add_argument(
'--delete-faulty-chunks',
action='store_true',
help='Try to delete faulty chunks after they have been '
'rebuilt elsewhere. This option is useful if the chunks '
'you are rebuilding are not actually missing but are '
'corrupted. '
'(default=%s)' % self.tool_class.DEFAULT_TRY_CHUNK_DELETE)
return parser
def _take_action(self, parsed_args):
if not self.distributed: # local
self.tool_conf['dry_run'] = parsed_args.dry_run
self.tool_conf['try_chunk_delete'] = \
parsed_args.delete_faulty_chunks
self.rebuilder = BlobRebuilder(self.tool_conf,
input_file=parsed_args.input_file,
logger=self.logger)
if self.distributed:
self.rebuilder.prepare_distributed_dispatcher()
else:
self.rebuilder.prepare_local_dispatcher()
for item, _, error in self.rebuilder.run():
if error is None:
status = 'OK'
else:
status = 'error'
yield (self.rebuilder.string_from_item(item), status, error)
def take_action(self, parsed_args):
if not parsed_args.input_file:
raise ValueError('Missing input file')
return super(ChunkRebuildCommand, self).take_action(parsed_args)
def test_rebuild_old_chunk(self):
for c in self.chunks:
convert_to_old_chunk(self._chunk_path(c), self.account,
self.container, self.path, self.version,
self.content_id)
chunk = random.choice(self.chunks)
chunk_volume = chunk['url'].split('/')[2]
chunk_id = chunk['url'].split('/')[3]
chunk_headers, chunk_stream = self.blob_client.chunk_get(
chunk['url'], check_headers=False)
os.remove(self._chunk_path(chunk))
chunks_kept = list(self.chunks)
chunks_kept.remove(chunk)
conf = self.conf.copy()
conf['allow_same_rawx'] = True
rebuilder = BlobRebuilder(conf, service_id=chunk_volume)
rebuilder_worker = rebuilder.create_worker(None, None)
rebuilder_worker._process_item(
(self.ns, self.cid, self.content_id, chunk_id))
_, new_chunks = self.api.object_locate(self.account, self.container,
self.path)
new_chunk = list(new_chunks)
self.assertEqual(len(new_chunks), len(chunks_kept) + 1)
url_kept = [c['url'] for c in chunks_kept]
new_chunk = None
for c in new_chunks:
if c['url'] not in url_kept:
self.assertIsNone(new_chunk)
new_chunk = c
# Cannot check if the URL is different: it may be the same since we
# generate predictible chunk IDs.
# self.assertNotEqual(chunk['real_url'], new_chunk['real_url'])
# self.assertNotEqual(chunk['url'], new_chunk['url'])
self.assertEqual(chunk['pos'], new_chunk['pos'])
self.assertEqual(chunk['size'], new_chunk['size'])
self.assertEqual(chunk['hash'], new_chunk['hash'])
new_chunk_headers, new_chunk_stream = self.blob_client.chunk_get(
new_chunk['url'])
chunk_data = b''.join(chunk_stream)
new_chunk_data = b''.join(new_chunk_stream)
self.assertEqual(chunk_data, new_chunk_data)
fullpath = encode_fullpath(self.account, self.container, self.path,
self.version, self.content_id)
self.assertEqual(fullpath, new_chunk_headers['full_path'])
del new_chunk_headers['full_path']
# Since we generate predictible chunk IDs, they can be equal
# self.assertNotEqual(chunk_headers['chunk_id'],
# new_chunk_headers['chunk_id'])
# We could compare the modification time of the chunks,
# but unfortunately they have a 1s resolution...
# self.assertNotEqual(chunk_headers['chunk_mtime'],
# new_chunk_headers['chunk_mtime'])
new_chunk_id = new_chunk['url'].split('/')[3]
self.assertEqual(new_chunk_id, new_chunk_headers['chunk_id'])
del chunk_headers['chunk_id']
del new_chunk_headers['chunk_id']
self.assertEqual(OIO_VERSION, new_chunk_headers['oio_version'])
del chunk_headers['oio_version']
del new_chunk_headers['oio_version']
del chunk_headers['chunk_mtime']
del new_chunk_headers['chunk_mtime']
self.assertEqual(chunk_headers, new_chunk_headers)